violated the model because of return migration, another 11 percent because of migration to the United States, and 20 percent had such a variety of patterns that they were assigned to the "other" category. It is noteworthy that 29 percent of all migrants came to Monterrey from a rural community with no intervening stops.
This is not the place for a full discussion of the appropriateness of the stage migration model, either as it applies to Monterrey or in general. What we are interested in here is its adequacy when applied to large cities; it may have a much better fit when applied to small or medium-size cities. There are, however, several reasons why we should not anticipate a good fit in many developing countries. One assumption of the model rarely made explicit is that the urban-size hierarchy is well developed. In effect, the model assumes a population distribution pattern similar to the Christaller "central place" model in which all parts of a given territory are equally habitable and there is a geometric arrangement of different sizes of communities so that any person randomly selected within the territory (excluding the perimeter areas) would be located the same distance from communities of different size as any other person. In other words, it is assumed that all prospective migrants have the same objective opportunity, as measured by direct-line distance, of access to communities of the next larger size class.
This assumption simply does not hold in many developing countries. The population distribution, due to geographical and historical reasons, is often very uneven and within the urban hierarchy itself there are many "holes." Consequently, many potential migrants would have difficulty in "finding" the properly sized community. Given the top-heavy urban-size hierarchy, reflecting the more rapid rate of growth of large cities as demonstrated by the Davis data (6), it is not surprising to find a good deal of direct rural-to-large-city migration. We may conclude: The lesser the level of urbanization of a country and the less developed the urban-size hierarchy, the less the conformity to the stage migration model
Rural vs. Urban Origins
We are still left with our problem of determining the relative importance of rural versus urban areas in providing migrants to large cities. In the case of Monterrey the community of origin (not place of birth but where respondent spent the greater part of his formative years between ages 5 and 14) of migrants by time of first arrival is as follows:
Rural (less than 5,000) 5 6%
Small urban (5,000 to 19,999) 21
Medium urban (20,000 to 99,999) 17
Large urban (100,000 and over) 4
Foreign 2 to allow for movement to any larger size class (a leapfrog pattern), 42 percent had partial conformity to the model. But this still leaves one half of the migrants who the rural interview information was obtained about persons absent from the village. For further methodological details of the two studies see Caldwell (12, Ch. 1) and Balan, et al. (1 3, Ch. 1). Substantively, Ghana differs from Mexico in the predominantly male rural-urban migration, in the multilingual character of the society, in the existence of polygamy, and